A first step in rationally treating disease is to assess the patient against a classification of diseases, the results being used to determine what kind of disease the patient has and to predict the person's response to various therapies. The effectiveness of the process depends on the quality of the classification. At least in the case of cancer, the advent of microarray methods to analyze DNA, RNA or proteins from tumor cells has started to refine and improve the classification of cancer cells. See, for example, Golub et al., 1999, Science 286, p. 531.
Further, van't Veer et al., 2002, Nature 415, p. 530, illustrates how such “molecular profiling” is improving cancer classification. Van't Veer et al. shows that the results of gene-expression profiling of breast tumors, carried out after they had been surgically removed, can be used to predict which patients will develop clinical metastasis (the spread of the tumor to other sites, where secondary tumors develop). Treatment for individual breast cancer patients is chosen according to various criteria, such as the extent of tumor spread (which involves determining tumor size), whether cancer cells have spread to the auxiliary lymph nodes and how many nodes are involved, and whether distant clinical metastases are present. In women with no evidence of metastasis, the mainstay of treatment aimed at curing the disease is the removal of the tumor and radiotherapy. Unfortunately some of these patients later develop clinical metastasis. Thus, there is a need to identify women who, after surgery, will require further (“adjuvant”) therapy for the microscopic deposits of cancer cells that may have already spread from the primary tumor. See, for example, Caldas and Aparicio, 2002, Nature 415, p. 484; and Goldhirsch et al. 1998, J. Natl. Cancer Inst. 90, p. 1601.
Adjuvant therapy uses pharmaceutical agents, such as oestrogen modulators or cytotoxic drugs that reach cancer cells through the bloodstream. Such treatments frequently have toxic side effects. Identifying women who might need such treatment has traditionally relied on various clinical and histopathological indicators (e.g., patient's age, degree to which the cancer cells resemble their normal counterparts, the ‘tumor grade’, and whether the cancer cells express the oestrogen receptor). Even taken together, however, these indicators are only poorly predictive. So, to save a sizable but small percentage of lives, many patients who would have been cured by surgery and radiotherapy alone go on to receive unnecessary and toxic adjuvant treatment.
The results of van't Veer et al., 2002, Nature 415, p. 530 as well as other studies are beginning to be used in classification schemes that attempt to characterize a biological specimen (e.g. tumor) from a patient into plurality of biological sample classes (e.g., breast cancer requiring adjuvant therapy versus breast cancer that does not require adjuvant therapy). A number of clinical trials, funded by companies and organizations such as the Avon Foundation, Millennium Pharmaceuticals, the European Organization for Research and Treatment of Cancer, and the National Cancer Institute, are presently underway to discover and validate such classification schemes. See, for example, Branca, 2003, Science 300, p. 238.
A number of biological classification schemes are available for breast cancer. For example, Ramaswamy et al., 2003, Nature Genetics 33, p. 49 provides a gene-expression signature that distinguishes primary from metastatic adenocarcinomas. Su et al., 2001, Cancer Research 61, p. 7388, describe the use of large-scale RNA profiling and supervised machine-learning algorithms to construct a first-generation molecular classification scheme for identifying carcinomas of the prostate, breast, lung ovary, colorectum, kidney, pancreas, bladder/ureter, and gastroesophagus. The Su et al., molecular classification scheme is useful in diagnosing metastatic cancers in which the origin of the primary tumor has not been determined. Wilson et al., 2002, American Journal of Pathology 161, provides an expression signature characteristic of HER2/neu positive tissue that is correlated with reduced survival of node-positive breast cancer patients. Richer et al., 2002, The Journal of Biological Chemistry 277, p. 5209, provides a genetic signature for human breast cancer cells that are over-expressing progesterone receptor-A (PR-A) and a genetic signature for human breast cancer cells that are over-expressing progesterone receptor-B (PR-B). As indicated by Richer et al., 2002, an excess of one or the other PR isoforms can result in tumors with different prognostic and hormone-responsiveness profiles from tumors that have equimolar levels of the two PR isoforms. Gruvberger et al., 2001, Cancer Research 61, p. 5979, provides a molecular classification based on DNA microarray data that can discriminate tumors based on estrogen receptor status.
The biological classification schemes outlined above are just a sampling of the many biological classification schemes that are available for breast cancer. Further, breast cancer, represents just one of many biological classifications of interest. Other representative biological classifications include a diagnosis of cancer generally and, even more generally, a diagnosis of a disease. One problem with each of these aforementioned biological classification schemes is that they each require specialized input (e.g., formatted microarray data). Thus, in an effort to characterize a biological specimen, the specialized input and output of each biological classification scheme must be deciphered. Because of such obstacles, medical care professionals typically use only a limited subset, at most, of such biological classification schemes.
Thus, given the above background, what is needed in the art are improved methods for making biological classification schemes available for classifying specimens into biological classes.
Discussion or citation of a reference herein will not be construed as an admission that such reference is prior art to the present invention.